Post on 23-Apr-2023
ARTICLE IN PRESS
0277-3791/$ - se
doi:10.1016/j.qu
�CorrespondE-mail addr
(R. Sinha), ng
(S.K. Tandon),
Quaternary Science Reviews 27 (2008) 391–410
Quaternary fluvial and eolian deposits on the Belan river, India:paleoclimatic setting of Paleolithic to Neolithic archeological
sites over the past 85,000 years
M.R. Giblinga,�, R. Sinhab, N.G. Royb, S.K. Tandonc, M. Jaind
aDepartment of Earth Sciences, Dalhousie University, Halifax, Nova Scotia, Canada B3H 3J5bDepartment of Civil Engineering, Indian Institute of Technology, Kanpur 208016, Uttar Pradesh, India
cDepartment of Geology, University of Delhi, Delhi 110007, IndiadRisø National Laboratory, Radiation Research Department, P.O. Box 49, DK-4000 Roskilde, Denmark
Received 18 April 2007; received in revised form 5 November 2007; accepted 5 November 2007
Abstract
Archeological sites in the bedrock Belan Valley at the southern margin of the Ganga Plains of India have unearthed Paleolithic to
Neolithic artifacts and the first known evidence for rice cultivation. We present a sedimentological and paleoclimatic analysis for Belan
sections, incorporating new luminescence and radiocarbon dates and a compilation of previous research. Some 20m of strata are exposed
in fluvial terraces, commencing with pedogenic and channel calcretes linked to groundwater ponding on the underlying bedrock.
Overlying alluvium deposited from mixed-load meandering rivers yields dates between 85711 and 7278 kyr before present (BP),
implying sustained fluvial activity during Marine Isotope Stage 5 and later; these strata contain Middle Paleolithic artifacts. Thin
reworked gravels with Upper Paleolithic artifacts are dated at �21–31 kyr BP, and may represent declining alluviation and floodplain
gully erosion during reduced monsoonal activity around the Last Glacial Maximum (LGM).
Younger channel fills contain shell-rich eolian sand, and mounds of shelly sand lie inland from the river. Five OSL dates from the
sands span 14 to 7 kyr BP, corresponding to a period of climatic instability that includes the Younger Dryas as the monsoon intensified
following the LGM. Although suggesting more arid phases, the source-bordering eolian material has a small volume and the grains are
partially bleached, indicating local wind action. Overlying floodplain muds reflect renewed alluviation, after which the river incised
during peak monsoon flow.
The Mesolithic settlement of Chopani-Mando spans a period of reduced monsoon activity and climatic instability following the LGM.
Subsequent Neolithic settlements were probably established under stronger monsoon conditions suitable for the development of
agriculture. Mesolithic habitation may have ended when a nearby bedrock channel was abandoned as the reinvigorated Belan cut a new
course, along which Neolithic settlements were later established.
r 2007 Elsevier Ltd. All rights reserved.
1. Introduction
In a classic book entitled ‘‘The Beginnings of Agricul-ture’’, Professor Sharma and his colleagues (1980) docu-mented a remarkable series of archeological sites along theBelan Valley at the southern margin of the Ganga Plains ofnorthern India (Fig. 1A). These sites are important for
e front matter r 2007 Elsevier Ltd. All rights reserved.
ascirev.2007.11.001
ing author. Tel.: +1902 494 2355; fax: +1 902 494 6889.
esses: mgibling@dal.ca (M.R. Gibling), rsinha@iitk.ac.in
roy@iitk.ac.in (N.G. Roy), sktand@giasdl01.vsnl.net.in
mayank.jain@risoe.dk (M. Jain).
establishing events in the human colonization of India(Misra, 2001). The book documents in particular UpperPaleolithic to Neolithic sites with abundant artifacts, andsets out evidence for domestication of animals and some ofthe world’s earliest agricultural activity, including the firstknown rice cultivation.Important questions arise concerning the climate and
environment under which these early human settlementswere established. How intense was the Southwest IndianMonsoon during this period, what types of rivers wereactive, and did they dry up periodically or shift elsewhere? Inapproaching these questions, much depends on obtaining
ARTICLE IN PRESS
25°N
Deoghat
Belan
Tons
Gorma
Tundiyari
Seoti (W)
Delhi
Allahabad
N
0 10km
Nai
na
Ganga
KYamuna
Craton
Ch
BK
AH
HFT
KA
A
Seoti
82°
CM
M
C1
C2
D1
D2
0 2
K
B
NRoad Abandoned
channel
8E°2
(E)
km
Fig. 1. (A) Location map of Belan River and sites, Uttar Pradesh and Madhya Pradesh, India. Inset shows location at southern margin of Ganga Plains:
A ¼ Agra, AH ¼ Allahabad, K ¼ Kanpur, KA ¼ Kalpi. Rivers: Ch ¼ Chambal, B ¼ Betwa, K ¼ Ken. (B) Field sections and localities (box in A), with
sections in Williams and Clarke (1995) and excavation sites of Sharma et al. (1980) in parentheses: D1 and D2 at Deoghat Bridge; M ¼Mahagara (Section
4 and excavations); K ¼ Koldihwa (excavation area; no section reported here); C1 and C2 at Chillahia (downstream from Chillahia section);
CM ¼ Chopani-Mando (same area as Sections 8A–C and 9 and excavations); B ¼ Belan Main Section of Williams and Clarke (1995), no log provided by
them and no section reported here. Remote-sensing image from IRS, LISS III, 1 July 2001. (C) The Belan River close to the position of abandoned Belan
channel. At right is the Belan Main Section of Williams and Clarke (1995). The section is the cut face of a terrace with 14m of floodplain muds resting on a
2m calcrete which forms a resistant ledge just above the river gravels. The strata overlie Vindhyan bedrock. Kaimur Hills are visible to the south.
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410392
reliable descriptions of the strata, coupled with a chron-ological framework designed to constrain key events. Aseries of papers have addressed some of these questionswhile providing a chronological framework for the Belansites (Sharma et al., 1980; Williams and Clarke, 1984, 1995;Pal et al., 2004; Williams et al., 2006).
We build on this earlier body of work to describe thegeomorphic setting of the area using remote-sensingimages, and to provide a detailed stratigraphic andsedimentological record for the Belan sites based onphotomosaics of the river cliffs, measured sections (Fig.1B), and age dates tied to key events. In particular, weprovide the first evidence from these sites for climaticinstability following the Last Glacial Maximum (LGM), inthe form of interaction between eolian and fluvial systemsduring the period of early human settlement. The studybenefits from the broader context provided by our previousstratigraphic work on the southern Ganga Plains (Giblinget al., 2005; Sinha et al., 2006; Tandon et al., 2006).
Although the predominantly fluvial records that wepresent here have relatively low resolution compared withsome lake and marine records, they are significant becausefluvial deposits represent the overwhelming bulk of IndianQuaternary deposits and because the early inhabitants ofthe Belan valley lived along the river. Additionally, theBelan sites provide paleoclimatic information for theseasonal, humid southernmost Ganga Plains, for whichlittle data has previously been available, and supplementsthe important lake record at Sanai Tal (Sharma et al.,2004). The Belan drains northern Peninsula India, whichmay generally have received less monsoonal precipitationthan the Himalaya to the north (Thamban et al., 2001;Staubwasser and Weiss, 2006). The study area is thus apotentially useful adjunct to paleoclimate analysis fromhigh-precipitation Himalayan areas (Phadtare, 2000), orfrom arid NW India (Prasad and Enzel, 2006) which hasoften been considered a standard reference area fornorthern Indian climate records.
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 393
2. Study methods
Geomorphic mapping (Fig. 2A) was based on remote-sensing data (IRS, LISS III, dated 1 July 2001, with spatialresolution 23.5m) obtained from the National RemoteSensing Agency (NRSA), Hyderabad. Mapping was aidedby image-enhancement techniques including contrast en-hancement to reduce the effects of haze and spatialenhancement using filters such as gradient and Laplacian.Contrast-stretching FCC and NDVI maps were used toassist in demarcating geomorphic unit boundaries, andfield checks were carried out to confirm the unit disposi-tions. A digital elevation model (DEM) was created using105 elevation points identified on topographic sheets, withan elevation range from 93 to 437m. Of these points, 95
Fig. 2. (A) Geomorphic units of the Belan areas, based on interpretation of re
map, viewed from northwest, with strong relief between the Kaimur Plateau a
were used to prepare the DEM and 10 were used foraccuracy analysis. Contour lines from 140 to 400m weredigitized. For accuracy analysis, the difference in elevationbetween topographic sheets and the DEM was determined,and the root mean square error was calculated as 1.06.A georeferenced 3D view was prepared and the rasterimage of the study area was draped over it (Fig. 2B).Channel planform characterization followed methods set
out in Friend and Sinha (1993) using the satellite images,and additional morphometric data were computed using1:50,000 scale topographic sheets coupled with satelliteimagery. Sinuosity and braid-channel ratio was estimatedover 5 km reaches.Field investigations in 2004 and 2005 were guided by site
information provided by Williams and Clarke (1995);
mote-sensing images. (B) 3D-perspective image from the digital elevation
nd the Belan Valley.
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410394
however, some sites had degraded such that sectionscould not be usefully measured, and some new sectionswere discovered. The archeological excavations wereno longer accessible. Serial photographs of cliff sectionswere taken from boat and bank traverses, and photo-mosaics were constructed and tracings made (Fig. 3).Guided by the lateral coverage available, nine sections weremeasured, including four at Chillahia 2 site, where thestrata varied considerably along the outcrop. In view of thedisposition of isolated cliff faces along several kilometers ofriver bank, coupled with great variation in the lateralextent of individual beds, the results are presented asindividual sections, and no attempt has been made togeneralize a bed-by-bed stratigraphic column. Paleoflowmeasurements were made from three-dimensional cross-beds, flute casts, and lateral-accretion surfaces. Selectedbulk samples were analyzed for grain-size distributionusing a Malvern laser diffraction Mastersizer 2000, andgrain-size statistical parameters are derived from Folk(1974).
Samples for luminescence dating were collected in light-proof PVC pipes, and were marked on the photomosaics.Sand-sized quartz fractions (63–90, 90–180, and 180–250mm,as available) were extracted and treated with concentratedHF for 40min to remove the alpha-irradiated skin. OSLdates on quartz were obtained using the single-aliquot
Fig. 3. Photomosaic tracings: (A) Mahaga
regenerative dose procedure (SAR: Wintle and Murray,2006). Both single grains and single aliquots were used formeasurements. The samples successfully passed the usualtests for the application of SAR procedures, includingthe preheat and cutheat dependence of De estimates andthe ability to recover a laboratory-delivered dose from thesamples.Jain et al. (2005a) showed that the SAR protocol works
well only on samples or aliquots in which the OSL signal isdominated by the fast component. The normalized averageOSL decay curves from the natural sample and fromlaboratory-irradiated aliquots (10Gy) (n ¼ 24) of oneBelan sample (053422) are shown in Fig. 4A. The OSLsignal decays to a near-background level within 2 s ofoptical stimulation, and the shapes of the natural and labregenerated signals are identical. These decay-curve char-acteristics suggest that the OSL decay curve is dominatedby the fast component.Growth curves for individual aliquots and grains were
generated and fitted with a single saturating exponentialcurve to estimate the equivalent natural dose (Fig. 4B). Thesensitivity-corrected natural signals (Ln/Tn) were wellbelow the saturation level (Fig. 4B). The De values fromindividual aliquots were determined by interpolating theLi/Ti values (sensitivity-corrected natural OSL signals)onto the growth curve.
ra section and (B) Chillahia 2 section.
ARTICLE IN PRESS
20 22 24 26 28 30 32 34 36 38 400
2
4
6
8
10
12
Fre
quency
De (Gy)
0 20 40 60 80 1000
5
10
15
20
25
De (Gy)
frequency
0 4 8 30 40
0.0
0.2
0.4
0.6
0.8
1.0
Time of decay (s)
Norm
alis
ed c
ounts
/s
Natural decay
Regn. decay
Test dose decay
De (Gy)
0 100 200 300 4000
Lx/T
x
1
2
3
4
0
50
100
150
200
250
300
0 2 6
Disk
De (
Gy)
De=150±9
-2
0
2
4
6
10
Regenerative dose (Gy)
8
Lum
inescence (
Li/
Ti)
Natural
0 50 100 150 200
1 3 4 5 7
2 6
Fig. 4. Luminescence sample information: (A) OSL signal decay curve for 40 s of stimulation with blue diode (90% intensity). The signal rapidly reaches
near-background values within a few seconds (approximately 2 s). Both the natural and lab generated signals are identical. (B) Laboratory-inserted dose
plotted against luminescence (optical) signal, corrected for sensitivity change, for each aliquot of sample 053422. The growth of the signal fits well with a
single saturating exponential curve. Sensitivity-corrected natural signal (Ln/Tn) is arrowed, and plots well below the saturation level. The error bar
represents standard error due to counting statistics. (C) The equivalent dose distribution of sample 053422 plotted as a histogram and as a radial plot. The
distribution is asymmetric and strongly skewed towards highest values. (D) The distribution of equivalent dose after completely bleaching sample 053422
under blue light at room temperature, and then employing a dose from an artificial source. The distribution is typical Gaussian. (E) Laboratory-inserted
dose plotted against luminescence (isothermal), corrected for sensitivity change, for a single aliquot of sample 053425. The natural ITL is plotted as a solid
black dot on the vertical axis. The error bar represents standard error due to counting statistics. The error indicates the interpolated dose equivalent to the
natural ITL. (F) The equivalent dose estimation of individual disks (aliquots) plotted with respect to the disk number. The 1s uncertainty shows that they
all are in overall agreement.
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 395
Equivalent dose distributions in histograms and radialplots for individual samples are asymmetrically distributed,and the distribution is strongly skewed towards highervalues (Fig. 4C). However, for a completely bleachedsample the dose distribution is Gaussian, and the radialplot shows that all data points of estimated doses fallwithin a 2y range of uncertainty (Fig. 4D).
Based on these observations, all samples were partiallybleached at the time of deposition, and the paleodose wasestimated using single-grain measurements (Table 2). Thedetails of these investigations will be reported separately.Dose rates were measured using high-resolution gamma
spectrometry with a detector resolution of 2 keV. Otherpre-treatments to minimize overestimation of dose rate
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410396
included sealing of samples to avoid 222Rn loss, heating toavoid absorption of radiation by organic matter and water,and grinding to o2 mm for homogenization. An internaldose rate of 0.0670.03mGy a�1 for quartz grains and adepth-dependent cosmic ray dose rate were also included.The dry dose rate was thus computed for these relativelycoarse grains and corrected for the ambient moisture.
Three samples in older strata were in OSL saturation forquartz. These samples were dated by isothermal thermo-luminescence (ITL) using a SAR protocol described byJain et al. (2005b). A number of thermal traps are presentin the quartz crystal (Aitken, 1985), and the study utilizedthe 325 1C dating trap with constant heating at 310 1C for250 s. The ITL signal grows smoothly with regenerationdoses and there is excellent reproducibility (within 5%of unity). The sensitivity-corrected natural ITL signalsintercept the growth curve well below the saturation level(Fig. 4E). About six aliquots were used for individual oldersamples. There was a good overall agreement betweenthe doses obtained from individual aliquots of the sample(Fig. 4F).
AMS radiocarbon dates were obtained on shell frag-ments at Geochron Laboratories, Cambridge, MA, USAusing the lab’s standard analytical methods. The resultswere provided in radiocarbon years before present (BP),and were calibrated using the software programme Calib5.0.2 (Reimer et al., 2004). They are reported as calendaryears BP with 1s error estimates.
3. Geomorphic setting of Belan Valley
The Belan River occupies a low-relief valley cut intoProterozoic quartzites of the Vindhyan Group (Fig. 1C)about 80 km southeast of Allahabad. The river runswestward parallel to the Kaimur Hills, a prominent uplandof the northern Indian Craton that lies a few kilometers tothe south (Fig. 2A and B). The Belan joins the Tons River,which flows northward to the Ganga Plains south ofAllahabad. Within the valley, up to 22m of Quaternaryalluvium is preserved as terraces on both sides of thechannel, resting on bedrock.
North of the Belan River, a gently undulating bedrockplateau with a thin soil and alluvial cover varies from 93 to116m above mean sea level in the study area. South of theriver, the bedrock rises gently below a thin alluvial blanketto the scarp front and top of the Kaimur at an elevation of437m. Two isolated hills (elevation 365 and 354m) lie closeto the study area, and several incised streams descendsteeply from the Kaimur to become tributaries of theBelan, which receives additional groundwater seepagedownstream. Present-day rainfall is 800–1000mm, risingto 1000–1400mm over the hills, mainly falling during theJune–September period of the Southwest Indian Monsoon.
Six geomorphic units are recognized in the area(Fig. 2A). The Tons Channel and Floodplain represents asingle-thread, low sinuosity (o1.32) bedrock channelwithout braid bars, flowing throughout the year and with
a narrow active floodplain. In reaches closer to Allahabad,the river is more sinuous and has large sandy bars with 2Ddunes. Minor Channels and Floodplains includes the Belanand its tributaries, also single-thread channels with narrowfloodplains. The Belan valley is a few hundred meters widebetween Quaternary terraces and, at low stage, has a graveland sand bed with shells and extensive weed cover.Sinuosity is mainly o1.5, with values of 1.56 and 1.89 intwo upstream reaches. For both Tons and Belan, olderalluvium forms prominent terraces that border the activefloodplain and are rarely flooded; exposures through theseterraces provide the bulk of our sections. Minor ephemeralchannels bring water to the trunk channel during themonsoon season. Several important archeological sites lieclose to a partially abandoned Belan channel (Fig. 1B),which still takes some flood water. Inactive Floodplains
have similar elevations to active floodplains, but borderabandoned channels and are not regularly flooded. Theyare characterized by dry channels, sparse waterloggedareas, and vegetation. The Belan and its tributaries havedrainage densities up to 2.45 km/km2—more than doublethat of the Tons and its tributaries but much lower thanalluvial rivers (Kale and Gupta, 2001). Very large floodsare not expected in a basin of this type.
Slightly dissected areas include the low-lying hills northof the Belan, where drainage density is low. They appearfeatureless, with sparse vegetation, waterlogged areas, andtracts of exposed rock. Some areas of the Kaimur Plateauto the southwest have a similar appearance. Highly
dissected areas lie south of the Belan, where the alluviumhas high-density gully networks, and much of the upland isalso heavily dissected. A unit of Plateau Scarp and Bedrock
Outliers represents the steep north-facing scarp of theKaimur Hills at an elevation of 160–300m, with densevegetation, moderate slope (up to 301), and low drainagedensity.
4. Facies descriptions
Six sections were described (Fig. 1B), and seven facieswere identified (Table 1) based on information fromphotomosaics (e.g., Fig. 3) and measured logs.
4.1. Channel deposits
These comprise Channel-base Gravel and Sand (Facies 1)and Laterally Accreted Sand and Mud (Facies 2). AtMahagara (Fig. 3A), a channel deposit 8m thick com-mences with a channel-base sheet of gravel and sand thatcomprises a single set of planar cross-strata, depositedfrom a 2D dune (Fig. 5A and B). The gravels includeVindhyan bedrock fragments and reworked carbonate andferruginous nodules from associated floodplain muds.Overlying trough cross-beds of pebbly sand (former 3Ddunes) pass updip into inclined sets of interbedded sandand mud with some rhizoconcretions, rising to 4m abovethe top of the gravel sheet. Paleoflow based on cross-set
ARTICLE IN PRESS
Table 1
Facies in Quaternary sections along the Belan River
Facies Sedimentary features and form Fossils and artifacts Interpretation
1.
Channel-
base gravel
and sand
Bedrock sandstone (to 7 cm) and reworked
carbonate and ferruginous (to 3 cm) clasts, as
unitso3m thick. Cemented where close to bedrock.
Forms: (A) boulder to pebble gravel (80 cm max.) as
thin lags on bedrock; (B) planar cross-sets up to
1.5m thick and co-sets, with inclined layers of
pebble gravel and sand; (C) trough cross-sets in
gravelly coarse sand, 1.5m cosets; (D) scour-based
lenses and irregular sheets of sand and gravel, with
flute and groove casts and sand shadows behind
pebbles.
Rare bivalve fragments and single valves in
current-stable position. Rhizoconcretions
common.
Bedload deposits and lags in lower parts of
channels, especially where narrow channels cut
cohesive floodplain deposits.
Form: Sheets and lenses at base of channel bodies,
or fill hollows cut into floodplain muds and valley-
base calcrete (Facies 3 and 6).
2. Laterally
accreted
sand and
mud
Inclined bedsets 1.5–7m thick of mud and very fine,
ripple cross-laminated sand o30 cm thick, dips up
to 201. Sheets planar to slightly sigmoidal, with tops
locally truncated. Rare inclined gravel sheets
(sandstone and reworked carbonate) o20 cm thick.
Sand commonly cemented with carbonate, and
carbonate nodules prominent, especially at tops.
Individual dipping layers extend �60m in downdip
direction.
Rhizoconcretions common. Bed- and suspension-load deposits of point bars
in meandering rivers. Parent channels o7m deep
and 100m wide, based on vertical and downdip
extent of inclined layers. Mainly low-energy flows.
Form: Inclined bedsets form sheets 4200m wide,
measured in downdip direction. Inclined layers pass
downdip into Facies 1, or rest directly on eroded
floodplain mud and calcrete (Facies 3 and 6).
3. Mud
with
carbonate
nodules
Sheets of structureless silty clay, buff to red-brown,
strongly cohesive, up to 8.5m thick. Weakly
stratified where silt and sand layers present.
Carbonate nodules sparse to abundant. Where
recognizable, component units a few meters thick
comprise paler mud with sparse carbonate, passing
up into darker, carbonate-rich mud. Thin sheets of
reworked ferruginous pebbles where muds rest on
bedrock.
Rhizoconcretions common. Archeological
sites at CM, K and M lie within topmost
0.5–2.6m of these strata.
Floodplain deposits with some stacked soil
profiles, or fills of abandoned channels and
scours. Modified by pedogenesis and vegetation.
Commonly overlie channel deposits in terrace
cuts.
Form: Sheets hundreds of m in extent, and lenses
above laterally accreted sheets (Facies 2) and against
channel cutbanks. Thin lenses above channel-base
sands.
4. Gravel
sheets in
mud
Reworked carbonate and ferruginous pebbles up to
2 cm diameter, above erosional surfaces. A few
centimeters to 1.5m thick.
Bivalve and gastropod fragments and
complete valves. Cemented Gravel IV (M)
with animal bones and stone tools (Sharma
et al., 1980).
Nodules from floodplain muds reworked on
gullied surfaces. Common in upper terrace cuts,
may be associated with gully erosion and local
channel flow.Form: Sheets and lenses within and overlying
floodplain deposits (Facies 3).
5. Massive
shelly sand
and silt
Buff, cohesive fine- and very fine-grained sand and
silt with clay matrix, up to 5m thick. Mainly
unstratified, rare thin layers of carbonate and
ferruginous pebbles to 2 cm diameter.
Dispersions and lenses of bivalve and
gastropod fragments and complete valves.
Eolian deposits within and adjoining channels.
Sediment probably blown from nearby channels
and ponds.
Form: Lenses o5m thick and 30m wide, above
channel Facies 1 and 2. Also meter-scale mounds in
belt �1 km wide on N side of Belan River.
6. Calcrete Coalesced nodules a few cm to dm in diameter, with
vertical and horizontal rhizoconcretions up to 1m
long. Drab mottles. Typically 1–3m thick.
Rhizoconcretions abundant. Stage IV calcrete (Machette, 1985), in channel
and floodplain deposits, especially at valley bases.
Local channel calcretes.
Form: Sheets and lenses above bedrock at most
localities. Also locally at tops of channel bodies and
within floodplain muds.
7. Gully
fills of mud
and
carbonate
Soft pale to dark mud with reworked carbonate
fragments a few cm in diameter. Massive to weakly
stratified. Up to a few m thick.
Charcoal and animal bones. Reworked
stone tools and fragments of pottery and
brick.
Partial fills of gullies, including cultural materials
at archeological sites.
Form: Lenses within modern gullies.
Archeological sites: CM ¼ Chopani Mando, K ¼ Koldihwa, M ¼Mahagara.
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 397
ARTICLE IN PRESS
Fig. 5. Photographs of outcrop sites, Belan Valley: (A, B) Channel body in the Mahagara section (Fig. 3A). Channel base is overlain by a sheet of planar-
cross-stratified gravel (p). Above is a trough cross-stratified pebbly sand (t) which passes updip into gently inclined sand and mud layers (LA, lateral-
accretion surfaces). Diffuse, thick beds of massive silt and sand with shell fragments (s), of eolian origin, mantle the accretion surfaces. A sheet of
floodplain mud (m) with carbonate nodules caps the section. Topmost carbonate gravel is just beyond cliff top. (C, D) Calcretes at base of section at
Chillahia 1 section. (C) Pedogenic calcrete �5m above section base. Continuous zone of coalesced nodules and rhizoconcretions (c) overlies mudstone
with carbonate nodules (m), with lenses of reworked carbonate nodules (rc). Scale is 75 cm long. (D) Channel deposits of cemented gravel �3m above
section base, composed of reworked carbonate fragments with planar cross-strata (p). Scale is 50 cm long. (E) View from Chopani-Mando archeological
site. Narrow bedrock valley �3m deep is cut into Vindhyan quartzites, with settlements on thin alluvial cover on adjacent terrace above bedrock. (F)
Modern Belan valley just east of Mahagara section (M). Recently aggraded inset terrace (it) borders older, higher terrace (t) farther from river. Asterisk
shows position of abandoned channel where it rejoins the Belan.
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410398
orientation was along strike of the inclined surfaces,confirming that the surfaces formed by lateral accretion.
At Chillahia 2 (Fig. 3B), a channel body 7m thick hassimilar lateral-accretion sets of sand and mud that arelocally scoured and extend into scour hollows along anerosional base (e.g., 10 and 70m positions). Individualaccretion surfaces can be traced for 25–35m downdip, andthe accretion set as a whole can be traced for more than100m in the downdip direction. The coarser beds containsparse ripple cross-lamination, and flute and groove castsand sand shadows are present in the basal strata, orientedobliquely with respect to the accretion surfaces. Overlying
muds fill an abandoned channel. Many sand layers—especially those low in the channel deposit—are partiallycemented with carbonate, probably from groundwateraction (Sinha et al., 2006).The lateral-accretion surfaces indicate that the channels
were meandering with well-developed point bars. For thetwo bodies illustrated in Fig. 3, the vertical extent ofchannel-base sheets and accretion surfaces suggests that thechannels were �7–8m deep, and the downdip extent ofindividual accretion surfaces (typically about two-thirdsof the channel width: Allen, 1965) suggests channel widthsof 40–50m. A third channel body with lateral-accretion
ARTICLE IN PRESS
BelanRiver (50m)
190 010
SOIL
+ Radiocarbon date
OSL date
0
5m
0 5
Clay+silt
Pebble gravel
Sand+silt
Silt
Kankar
Bivalve+gastopod shellsand fragmentsCross beds
DG-1/4-L8.28±0.51
DG-1/4B-S10.01±0.14
DG-1/4A-S13.03±0.09
m
Fig. 6. Field sketch at Deoghat Bridge, Log 1. Section is exposed in gullies on west side of road and 50m inland from the bridge.
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 399
surfaces at Chillahia 1 is 3m thick. The interbedded sandand mud in the accretion sets implies seasonal flowconditions, a high suspended load, and modest flowstrength, although flows were capable of mobilizing graveland scouring accretion surfaces. Banks of cohesive mudand underlying calcrete impeded channel erosion.
At Deoghat Bridge 1 (Fig. 6), 4m of gravel is present in asmall exposure. The basal deposits are structureless and areoverlain by stacked metre-scale planar cross-sets of gravelwith shells. These stacked deposits of 2D dunes probablyformed on a point bar, possibly as inwardly directed scrollbars bordered by a landward swale.
4.2. Floodplain deposits
These comprise Mud with Carbonate Nodules (Facies 3)and Gravel Sheets in Mud (Facies 4). Facies 3 forms thebulk of the outcrops, as sheets up to 8.5m thick that extendalong cliff faces for 4300m at Chillahia 2, the mostcontinuous section. The muds are structureless, buff tobrown silty clay, with carbonate nodules and rhizoconcre-tions (kankar), drab and dark mottles, and a few sheets ofsilt and sand. They represent the progressive accumulationof fine detritus from overbank flooding, probably out ofthe associated meandering channels. The scarcity of coarserflood layers suggests that overbank floods rarely tapped thecoarser bed load, and no finely interstratified sand and mudsheets suggestive of levee deposits were identified. A fewdiffuse, metre-scale layers with a higher proportion of
carbonate suggest enhanced pedogenic activity (Sinhaet al., 2006), although no distinct soil horizons wereidentified. The abundance of carbonate nodules andrhizoconcretions suggests seasonal conditions and a wellvegetated floodplain.Facies 4 is represented at several outcrops by erosionally
based sheets of carbonate gravel a few centimetres to 1.5mthick, capping terraces or enclosed within mud sheets. Theyrepresent floods that eroded kankar from the associatedmuds and redeposited them on gullied floodplain surfacesalong with bivalve and gastropod fragments, animal bonesand stone tools (Sharma et al., 1980). At Mahagara, thehigh topographic level of the gravel sheet (top right ofFig. 3A) and scarcity of sand and bedrock gravel is not inaccord with floods from the main channel. Rather, wesuggest that these layers mark episodes of gully erosion ona degrading floodplain (terrace top), during whichreworked kankar was transported by monsoonal flows.Similar gravel-filled channels and larger gully fills ondegraded terrace tops are common in the southern GangaPlains (Gibling et al., 2005).
4.3. Eolian deposits
Massive Shelly Sand and Silt (Facies 5) was identified atthree localities. Up to 5m thick, these distinctive units ofunstratified buff sand and silt contain an abundance ofscattered shell fragments, locally concentrated into lenses,with a small proportion of complete shells.
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410400
Two occurrences are within channel fills. At Mahagara(0–40m position: Fig. 3A) lateral-accretion sets of sandand mud are overlain by more than 4m of massive shellysand and silt with diffuse, widely spaced inclined surfaces(Fig. 5A and B). At Deoghat Bridge 1 (Fig. 6), 4m of sandwith fragments and whole shells of bivalves and gastropodsforms a lens associated with two planar-cross-stratifiedgravel beds. The sand is divided into two units by a thingravel layer that extends from the top of the lower cross-set, indicating two depositional periods of both gravel andsand.
The third occurrence (Deoghat Bridge 2) lies near thenorth bank of the Belan River, about 200m east of column1. In this area, low vegetated mounds several meters highform a 1 km belt that dies out away from the river. Onemound near the river, excavated by villagers, is 3m highand consists of buff sand with shell fragments, similar tothe material in the channel fills. Thin layers of reworkedcarbonate and ferruginous nodules are present. Moundsfarther inland are not exposed, but they appear to rest onfloodplain muds which are locally exposed below themounds.
The mean grain size of four samples from the Deoghatlocalities ranges from 2.81 to 3.61+ (fine- to very-fine-grained sand), the sorting parameter from 2.65 to 3.35(very poorly sorted), and all samples are positively skewed.Clay comprises 10–13% of the samples.
We interpret the unstratified shelly sand and silt as eoliandeposits blown into channels and swales or moundedon flat-lying, elevated floodplain surfaces. The sedimentwas derived from the deflation of dried-up channels andponds, as indicated by the abundance of shell fragments(shells are common in the modern river). Furthermore, thelocation of the mounds close to the modern Belan channelsuggests a nearby fluvial source, an inference supported bythe sand-sized sediment and the relatively poor sorting.The positive skewness suggests the addition of some finedust.
Williams and Clarke (1984, 1995) described from severalsections in the Belan and Son valleys massive clay loamsand fine sandy clays with pedogenic carbonate nodules,which they interpreted as loess. Although the strata theydescribed may include the deposits interpreted by us aseolian at Mahagara, they also include overlying muds, thatwe interpret as floodplain deposits (Facies 3) by compar-ison with many exposures in the southern plains (Giblinget al., 2005). These thick, weakly stratified depositsfrequently have a buff-coloured, case-hardened surfacewhich gives them a resemblance to loess. However,excavation reveals that they contain abundant clay, mottles,carbonate nodules, and rhizoconcretions, in accord with afloodplain origin and soil development. The high claycontent contrasts strongly with the low clay content of ourBelan eolian deposits. Although carbonate nodules arepresent within eolian silts and sands elsewhere in India(Tandon et al., 1997), the clay-rich nature and soil featuresof the Belan deposits strongly support a floodplain origin.
4.4. Calcrete
Calcrete (Facies 6) rests on bedrock and is welldeveloped in the basal 6m of the Chillahia 1 section andat the base of Chillahia 2 (Fig. 3B). It also forms a massive,2m cemented unit at the base of the Belan Main Section(Fig. 1C) of Williams and Clarke (1995). Calcrete is notdeveloped within the bedrock. At higher elevations withinterrace deposits, thinner calcretes are present at channeltops and within floodplain deposits.The calcretes consist of coalesced carbonate nodules with
interspersed mud and a faint vertic fabric (Fig. 5C).Vertical to horizontal rhizoconcretions are up to 1m long.At numerous localities (e.g., Chillahia 1: Fig. 5D), thecalcretes are associated with channel fills up to 2.5m thickcomposed of cemented, cross-stratified carbonate gravel.The cross-strata are curved in planview, and form part ofsmall lobate gravel bars, with fragments eroded from thenearby calcrete. Reworked lenses of carbonate gravellocally cap the calcretes.The nodular form, evidence for roots, and intraforma-
tional erosion suggest that the calcretes are largelypedogenic and have variably attained Stage II–III (nearcoalescence) and Stage IV (coalesced nodules) of develop-ment (Machette, 1985). Similar calcretes are common infloodplain deposits across the southern Ganga Plains(Sinha et al., 2006) as well as in northwest India (Dhir etal., 2004). The position of many Belan calcretes on or justabove bedrock suggests comparison with the channelcalcretes described from Spain by Nash and Smith(2003). These precipitate at depth in porous bedrock gravelwhere bicarbonate-charged groundwater flows laterallyabove impermeable bedrock. The massive, cementedcalcretes at the base of the Belan Main Section areinterpreted as channel calcretes, but other Belan calcretesare clearly pedogenic or reworked into channels, althoughtheir cementation may have been aided by groundwaterponding on underlying bedrock.
4.5. Other facies
Terrace tops are extensively gullied, and some gulliescontain colluvial material of soil, artifacts (includingpottery and brick), and animal bones (Gully Fills of Mud
and Carbonate: Facies 7).At Chillahia 2, parallel sets of clastic dykes a few
centimeters wide consist of cemented sand and penetrateup to 5m of strata (Fig. 3B). Similar features elsewhere inthe Ganga Plains have been attributed to earthquakeactivity (Agarwal et al., 2002; Gibling et al., 2005).
5. Paleoflow
Five planar and trough cross-sets and one occurrence ofsand shadows associated with flute casts, collectively fromfour sections, yield W to NW paleoflow directions(240–3501)—broadly parallel to the present Belan course.
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 401
One additional cross-bed yielded a flow direction of 0601.Lateral-accretion surfaces (Fig. 3) show considerablevariation in their strike, in accord with a meandering-channel planform.
6. Stratigraphy and chronology
6.1. Reliability of age dates
The chronology presented here is based on OSL, ITL,and radiocarbon dates. Table 2 brings together 33 agedates for the Belan sites, obtained using several methods atnumerous sites over a prolonged period, with notes ondates that have been used to define archeological periods.Our 13 new dates are highlighted in the table, and arediscussed below in a sedimentological context, before weaddress broader questions with the addition of dates fromprevious research.
A crucial question is the comparability of dates obtainedusing different methods, given that some of the datesdiscussed here—including dates from shells—have beenused to constrain the age of archeological sites. At someGanga Plains sites, quartz OSL and shell radiocarbon datesfrom the same beds are in close agreement (Gibling et al.,2005), but other sites show considerable discrepancy (N.Roy, unpublished data). At Mahagara, shell and OSLdates match reasonably well (Fig. 3A), but shell dates atDeoghat are several thousand years older than the OSLdates (Fig. 6). Older shell dates may reflect reworking ofpreviously deposited shells, and there is precedence fromsurface marine shell accumulations for reworking of shellsseveral thousand years old (Martin et al., 1996; Resig,2004). Alternatively, the shells may have drawn theircarbon from groundwater depleted in 14C flowing throughbedrock carbonates or older calcretes. For a Tibetan lake,Ji et al. (2005) inferred a reservoir effect of 1039 years. Incontrast, Owen et al. (2007) found that OSL datesappeared to be anomalously young relative to radiocarbondates on shells in the same deposit, although it was notclear why.
We provisionally base our assessment on the OSL andITL dates. Comparison of OSL and ITL dates for the samesamples shows good agreement for one Ganga Plains site(Jain et al., 2005b). Williams et al. (2006) presented datesfor Belan sections based on the IRSL method, which usesfeldspars. Although no tests were carried out to explorelong-term fading, these authors also noted a generallyreasonable correspondence with calibrated radiocarbondates on shells and charcoal, and their dates broadly matchour ITL dates for the older sections, although the sampleswere not collected from the same beds.
6.2. Studied sections and dates
Sharma et al. (1980) presented a single stratigraphicsuccession for the area, but Williams and Clarke (1995)presented a series of individual sections, noting that many
beds are discontinuous. We similarly present here informa-tion for individual sections (Fig. 7) before presenting ageneralized column (Fig. 8) which should not be taken toimply that such an apparently continuous section existsanywhere.The two Chillahia sections (22 and 15m thick: Fig. 7)
consist predominantly of stacked units of floodplain mudswith pedogenic features. One meandering-channel body ispresent in each section, and calcretes overlie the bedrock atboth localities, with fragments reworked into smallchannels at site 1. Three ITL dates for the channel bodiesrange from 7278 to 85711 kyr BP. The underlying mudsand calcretes are undated. A gully fill with artifacts andcharcoal was noted at the section top, but was not dated.The Mahagara section, 12m thick, contains a mean-
dering-channel filled with shelly eolian deposits. Three OSLdates for the Mahagara eolian sediments range from9.0970.9 to 12.9571.24 kyr BP. The date low in thechannel fill (12m position in Fig. 3A) is the oldest of thethree, and the two dates from dipping beds at the 32mposition are younger and in stratigraphic sequence. Oneshell date agrees well with the topmost OSL date. DeoghatBridge 1 section, 4m thick, includes fluvial gravels andshelly eolian deposits. Deoghat Bridge 2 is an eolianmound near the bank of the modern Belan river. OSL datesat the two localities are 7.5170.49 and 8.2870.51 kyr BP,overlapping within a 1s error range. The shell dates fromthese sites are considerably older.The topmost gravel layer at Mahagara contains late
Upper Paleolithic artifacts and geometric microliths (Palet al., 2004), and yielded a shell date of 11.2570.05 kyr BP.This layer represents gullying and degradation of flood-plain muds at the terrace top, and post-dates the eolian fillof the abandoned channel. We note that the OSL date of9.09 kyr BP for the topmost eolian material (Fig. 3A) isconsiderably younger than dates for shells and charcoalfrom the overlying gravel (Table 2), which could have beenreworked from older deposits.A poorly exposed section above bedrock near the
Chopani-Mando excavation site commences with a 50 cmgravel rich in ferruginous fragments, overlain by 4m offloodplain muds with kankar. Much of this muddymaterial had been stripped off by erosion prior to stone-age occupation of the site.Individual terrace sections along the Belan River
represent different periods, suggesting that the floodplaindeposits formed in local reaches and were not system-wide.The youngest river terraces are commonly banked againstolder terraces within embayed areas (Fig. 5F). Severalterraces with different ages show a broadly similar upwardtransition from calcrete above bedrock to alluvium(with channel bodies just above the modern Belan levelat two localities), to capping floodplain muds withreworked carbonate gravels. This similarity reflects pro-gressive aggradation of floodplains to a level beyondregular reach of Belan flooding, followed by degradationand incision.
ARTICLE IN PRESS
Table 2
Compilation of age dates for Belan sections
Section Source Lab no. Material and
method
Luminescence date Radiocarbon date Comments
Dose rate
(Gy/kyr)
Dose (Gy) Date (kyr) Date
(kyr BP)
14C date, BP
Mahagara
index trench
3 PRL 409 Charcoal NA 3.2–4.0 Midden with
Neolithic
artifacts at
lower level
behind
outcrop.
3 PRL 408 Charcoal 3.25–3.85
3 PRL 407 Charcoal 3.4–3.8
3 BAI/MGR
77-1
Charcoal 3.4–4.0
Koldihwa
excavation
3 �PRL 101 Charcoal 6.95–7.7 Associated
with
Neolithic
artifacts.
3 �PRL 100 Charcoal 7.65–8.75
3 �PRL 224 Charcoal 8.95–10.25
Deoghat
bridge 2
1 GX-31787-
AMS
Shell 11.1870.05 9740760 Eolian
mound on
river bank.1 DG-2/1-L OSL 2.0870.18 29.270.52 7.5170.49
Deoghat
bridge 1
1 GX-31786-
AMS
Shell NA 10.0170.14 9000760 Fluvial and
eolian
sediment in
channel fill.
1 GX-31785-
AMS
Shell 13.0370.09 11,1407100
1 DG-1/4-L OSL 1.7270.16 23.2870.86 8.2870.51
Mahagara 1 GX-31789-
AMS
Shell NA 11.2570.05 9840760 Carbonate
gravel at cliff
top
(Cemented
Gravel IV of
Sharma
et al., 1980).
Associated
with
Microlithic
artifacts;
predates
Neolithic
site.
3 +SUA 142 Shell 11.2–12.0
3 PRL 602 Charcoal 12.93–13.4
3 PRL 603 Charcoal 15.75–18.55
Mahagara 1 MH-1/6-L OSL 1.5570.15 20.1770.28 9.0970.90 Eolian
sediment in
channel fill.
1 MH-1/5B-L OSL 1.2370.12 17.8970.47 12.3071.19
1 MH-1/5A-L OSL 1.4470.13 21.7070.32 12.9571.24
1 GX-31788-
AMS
Shell NA 9.6470.09 8700760
Deoghat
bridge (7)
3 Beta 4789 Shell 20.85–22.05 Gravel layers
within or
below fine-
grained
(floodplain)
units.
Koldihwa 3 *TF 1245 Shell 22.55–24.45
Koldihwa
(7D)
3 Beta 4877 Shell �28–30
Deoghat 3 *PRL 86 Shell �28–30
Locality
unknown
3 Beta 4790 Shell �29–31
Chillahia 1 1 NCH-1/4-L ITL 2.8670.28 20779 7278 Meandering
river
channels
with lateral
accretion
surfaces.
Chillahia 2 1 CH-2/2a-L ITL 3.0770.31 246718 80710
1 CH-4/3b-L ITL 1.7870.18 152713 85711
Belan main
section (B-1)
and section
100m to NE
(other
samples)
2 B-8 IRSL 1673 Silt in gravel/
sand unit.
2 B-7 IRSL 2073 Silt-
dominated
sediments.
2 B-5 IRSL 45712
2 B-3 IRSL 52710
2 B-1 IRSL 90720
Age columns show calibrated radiocarbon dates, OSL and ITL dates.Sources: 1 (shown in bold) ¼ this study and Roy et al. (submitted for publciation),
reported with 1s range; 2 ¼Williams et al. (2006); 3 ¼ Sharma et al. (1980), Williams and Clarke (1984, 1995), and Pal et al. (2004), with calibrated ages
reported in Williams et al. (2006) at 2s range, and with section numbers from Williams and Clarke (1995). Dates considered by Sharma et al. (1980) to
provide reliable dates for archeological levels: * ¼ Upper Paleolithic,+ ¼ predates Neolithic, � ¼ Neolithic. NA ¼ not applicable.
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410402
ARTICLE IN PRESS
Fig. 7. Summary stratigraphic logs (see Fig. 1B for locations, and Table 2 for age dates).
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 403
6.3. Geological history of Belan River sections in a regional
context
The Belan outcrops occupy an upland valley more than1000 km upstream of the tidal limit at the Ganga Delta,and it is unlikely that the Belan was affected by changes insea level. Because the Belan is a tributary of the Tons whichin turn drains into the Ganga, changes in the equilibriumprofile of the Ganga north of the study area might haveaffected the Belan’s profile. However, the Belan is cut intobedrock along much of its course, and such effects shouldhave been minimal. The area is not a region of currentseismic activity, although clastic dikes suggest occasionalpast activity. Thus, we interpret the Belan succession interms of paleoclimate proxies and local intrinsic changes tothe river course.
Fig. 8 presents a summary column with age dates fromour work and previous work (Table 2) alongside two longproxy records for the monsoon. Upwelling and SouthwestIndian Monsoon intensity is based on stacking of five proxyrecords from cores in the Arabian Sea (Clemens and Prell,2003), from which a considerable proportion of India’sprecipitation is derived. Prasad and Enzel (2006) noted thatupwelling is a response to monsoonal wind strength but maynot be a reliable indicator of actual monsoon precipitationover northern India. Consequently, a record for modeledmonsoon rainfall is also included (Prell and Kutzbach,1987), representing general trends for Asia.Much of the discussion below deals with the period
following the LGM of Marine Isotope Stage 2. Sites acrosssouthern Asia and in offshore areas record intensificationof the monsoon following the LGM, with peak monsoon
ARTICLE IN PRESS
-2 -1 0 1 -30 -10 10 30
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
0Fluvial terrace aggrades(inset)Eolian/Fluvialactivity
Meanderingchannels,floodplains+ soils
Calcretes(undated)
1
2
3
4
5
6
(May includeseveral eventsof local terraceaggradationand incision)
ThisStudy
Arabian SeaSummer MonsoonFactor
SW Indian MonsoonChange in Precipitation(%) for S. Asia
MarineIsotopeStages
Ka B.P.
VINDHYANBEDROCK
72±8
80±10
85±11
~7-11(n=10)
3.2-4(n=4)
PreviousStudies
90±20
52±10
45±12
~21-31(n=5)
16±320±3
Terrace incision+ gullying(3.4 - 10.3;n=3)
Neolithicartefacts
U. Paleolithicartefacts
M. Paleolithicartefacts
Late U. Paleolithic-Advanced Mesolithicartefacts
Fig. 8. Events recorded in Belan outcrops, based on age dates in Table 2. Comparison is drawn with Arabian Sea Summer Monsoon Factor, based on
stacked proxy ocean records, from Clemens and Prell (2003), and for modeled monsoon rainfall in Asia from Prell and Kutzbach (1987). Approximate
duration of marine isotope stages is also shown.
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410404
conditions in the early to mid Holocene, followed by adecrease in monsoon intensity. However, the precise timingof these events appears to vary locally. Overpeck et al.(1996) suggested that the Southwest Monsoon peaked inthe 10–5.5 kyr BP period, in accord with pollen recordsfrom Tibet (van Campo and Gasse, 1993). In contrast,several records from India suggest that the monsoonpeaked after 8–9 kyr BP and perhaps at 6 kyr BP or later.Evidence from Thar Desert lakes suggest that precipitationincreased to a maximum at �7.2–6.0 kyr BP (Prasad andEnzel, 2006), and Himalayan peat records suggest that themonsoon strengthened after 7.8 kyr to a peak between 6and 4.5 kyr BP (Phadtare, 2000). As a result of orographicprecipitation on the Western Ghats, Peninsular Indiaexperiences reduced rainfall and the monsoon may haveintensified in this region only after 8–9 kyr BP (Thambanet al., 2001; Staubwasser and Weiss, 2006). As noted by
Prasad and Enzel (2006), Asia receives precipitation fromboth the Southwest (summer) and Northeast (winter)monsoons, and lake and river systems are affected by localtemperature and evaporation, groundwater supply, andregional precipitation gradients. Additional climatic per-turbations arise from mid-latitude westerlies and El NinoSouthern Oscillation (Owen et al., 2005; Williams et al.,2006), and local topography and rain-shadow effects mayinfluence thresholds for response recorded in the proxyrecords.In view of the strong eastward increase in precipitation
down the modern Ganga Plains (Sinha et al., 2005), animportant site for climatic comparison is an 18,000 yrrecord from Sanai Tal (lake), laid down in a meandercutoff northwest of Allahabad (Sharma et al., 2004) andabout 150 km from the Belan sites. These authors inter-preted the pollen and isotope record in terms of lake size
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 405
and precipitation changes. The lake expanded at about15.0 kyr BP as a result of increased monsoon intensity, afterwhich a drier period between about 12.2 and 13.4 kyr BPmay correspond to the Younger Dryas event. The lakeenlarged again after about 11.5 kyr BP, reaching itsmaximum size at about 6.7 kyr BP. We also draw upondocumented responses to climatic changes at southernGanga Plains sites within a few hundred kilometers of theBelan (Williams and Clarke, 1995; Srivastava et al., 2003;Gibling et al., 2005; Tandon et al., 2006).
Building on the earlier analysis of Williams et al. (2006),we recognize a series of events, although the large errorbars of some dates and discrepancies in dates obtainedusing different techniques must be borne in mind.
(1)
Channel-base calcretes above the Vindhyan Unconfor-mity (MIS 5 or older): Although these beds areundated, dates from a few meters above calcretes attwo of our localities and the Belan Main Sectionconstrain their formation to about 100 kyr BP or older.The calcretes could represent a long period of forma-tion, and could also be of different ages in differentsections. Our evidence suggests a predominantlypedogenic origin, but channel calcretes in particularcould represent a range of ages.
(2)
Fluvial deposition (�85–16 kyr BP, MIS 5–2): Therecord for this �70kyr period consists entirely ofalluvium—predominantly muddy floodplain depositswith some meandering-river channel bodies. Our ITLdates indicate that channels were flowing in the72–85kyrBP period, and IRSL dates suggest floodplainbuildups through to 16kyrBP. The sand/mud alterna-tion in lateral-accretion sets implies relatively gentle flowsunder seasonal conditions, with little evidence for high-intensity floods, in accord with the low drainage densityof the modern Belan in its bedrock setting. A prolongedperiod of fluvial activity is in accord with generally highprecipitation levels in MIS 3–5 (Fig. 8), and with thewidespread occurrence of fluvial facies elsewhere in theGanga Plains during this period (Goodbred, 2003;Srivastava et al., 2003; Gibling et al., 2005).Williams et al. (2006) suggested repeated aggradationand incision during this period, linked to variation inmonsoon strength (Fig. 8) and changing balance betweensediment and water discharge. The systematic upwardchange in facies within many terraces along the Belanand elsewhere in the southern Ganga Plains (Gibling etal., 2005) supports this suggestion. Highly dissectedterrains such as those south of the Belan should havecaused rapid erosion and sediment supply, promotingaggradation. The difference in age between terraceremnants suggests that some aggradation and incisionevents were restricted to individual reaches. Because thepreserved record is biased towards periods of fluvialaggradation, evidence for different depositional modes(e.g., lakes, windblown dunes) at other times may not beapparent.
Although we have not obtained LGM dates, previousresearch recorded several dates in the 21–31kyrBPperiod from shells in thin, reworked gravel lenses,probably within floodplain muds. Williams et al. (2006)suggested that alluviation continued through this period.Although we have no sections at these localities, weprovisionally suggest that the reworked gravels recordrecurrent episodes of floodplain gullying and erosionduring a period of reduced monsoonal precipitationaround the LGM. Ganga river discharge was consider-ably reduced after about 27kyrBP (Gibling et al., 2005).Closer to the Belan, at Telauli on the Yamuna River westof Allahabad, thick floodplain deposits are capped by3m of stratified yellow silt with abundant gastropods,which yielded a calibrated radiocarbon date of17.15–19.85 kyrBP (Williams et al., 2006). We suggestthat this bed is of lacustrine origin, and represents asimilar drawdown in fluvial activity in a large plains rivervalley around the LGM.
(3)
Fluvial and eolian activity under climatic instability(14–7 kyr BP): The alternation of fluvial and eoliandeposits at Mahagara and Deoghat suggests a period ofclimatic instability between about 14 and 7 kyr BP,representing the range of OSL dates with 1s errorbars. The return to floodplain deposits at the topof the Mahagara section suggests subsequent wetterconditions.
The older eolian dates in the Belan may correspondwith the Younger Dryas phase between about 12.2 and13.4 kyr BP recorded at Sanai Tal. Elsewhere in theGanga Plains, gully erosion (implying reducedfloodplain activity) took place at Mawar in the9–13 kyr BP period (Gibling et al., 2005). Cool, dryconditions associated with the Younger Dryasevent have been widely identified in Asian continentaldeposits (Porter, 2001; Ji et al., 2005; Jin et al.,2005; Nakagawa et al., 2005; Yancheva et al., 2007).The younger Belan eolian dates are in accord with agradual intensification of the monsoon in India to apeak at about 6–7 kyr BP (Sharma et al., 2004; Prasadand Enzel, 2006). Cool, dry events are documented forthis period at sites across Europe and Asia (van Campoand Gasse, 1993; Ji et al., 2005; Jin et al., 2005;Staubwasser and Weiss, 2006; Weninger et al., 2006),some probably linked to an 8.2 kyr BP event of globalimportance.
Although eolian deposits suggest aridity, the Deo-ghat locality contains interbedded fluvial and eoliansediments, and the availability of loose sediment andshell material implies that river action providedsediment for eolian transport. The presence of source-bordering eolian material may imply enhanced fluvialactivity, perhaps consistent with an intensifying mon-soon. The partial bleaching of our eolian samples issurprising, given that eolian sediments are normallywell bleached through prolonged exposure to sunlight.The eolian sediments are poorly sorted, and we infer
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410406
that they were locally derived, rather than indicative ofregional aridity and long-distance transport. The localcontext of our deposits is also important to note. Thevolume of eolian material is small, and the valleysetting and proximity to the Kaimur Hills (Fig. 2)allows for local, transient wind effects. The Mahagarachannel may have been abandoned due to meandercutoff, rather than representing reduced discharge, andthe Deoghat eolian sands fill a local swale. On the otherhand, no similar eolian deposits were identified in thewidely exposed older Belan record, which appears to beentirely fluvial, and the eolian materials appear to berestricted to a narrow time interval. Thus, we infer thatthe eolian deposits indicate periodically drier condi-tions, even though enhanced river discharge may haveprovided the sediment.
(4)
Terrace incision at Mahagara (after 8–10 kyr BPperiod): Incision of this terrace postdates the formationof the terrace-top floodplain layer and reworked gravelsheet. As noted earlier, the 11.25 kyr BP shell date forthe gravel is a maximum age and, additionally, incisionhas cut through underlying eolian strata dated at9.0970.9 kyr BP. Incision is inferred to have takenplace after 9 kyr BP and perhaps considerably laterthan this. Williams and Clarke (1984) and Williamset al. (2006) suggested that Belan incision representsmonsoon intensification and increased fluvial energy, inaccord with dates that bracket recent incision acrossnorthern India (Srivastava et al., 2001; Tandon et al.,2006). Evidence from Sanai Tal suggests that themonsoon peaked at about 6.7 kyr BP, and much ofthe incision of floodplain deposits may have taken placeat about this time.
(5)
Inset terrace aggradation at Mahagara: Decreasingmonsoonal activity since about 6 kyr BP is documentedacross northern India (e.g., Srivastava et al., 2001) andis represented in precipitation models (Fig. 8). Decreas-ing fluvial energy would have promoted renewed fluvialaggradation (Williams et al., 2006).7. Implications of paleoenvironment and paleoclimate for
archeological sites
7.1. Dating of sites
Lower Paleolithic tools were reported from basal Belangravels, with numerous Middle and Upper Paleolithic toolsfrom gravels in younger strata, including Middle Paleo-lithic tools from sections near Chillahia (Williams andClarke, 1995; Williams et al., 2006). No occupation siteswere identified, but stone tool ‘‘factory sites’’ are presenton the slopes south of the Belan where quartzite bouldersare available, and have yielded Lower to Upper Paleolithicartifacts (Srivastava, 1982). Our three ITL dates between72 and 85 kyr BP from Chillahia appear to represent stratafrom which Middle Paleolithic discoveries were made, in
accord with evidence that the Indian Middle Paleolithicdates from at least 125 kyr to perhaps 40 kyrBP (Mishra,1995). The strata represent meandering rivers and muddyfloodplains, probably with seasonal flow.Upper Paleolithic artifacts were obtained from sections
at Koldihwa and Deoghat that yielded dates in the22–31 kyr BP range (Table 2). The occurrence ofimplements in thin layers of reworked gravel impliesreduced rates of floodplain deposition and periodicgully erosion during this period of regionally reducedprecipitation.Three major excavation sites were described by Sharma
et al. (1980). At Chopani-Mando, 1.55m of siltyclay with kankar were excavated just above bedrock on arock-cut terrace above the old Belan river course (Fig. 9A),and yielded a succession of late Upper Paleolithic toAdvanced Mesolithic (or Proto-Neolithic) artifacts. Theexcavations discovered huts, hearths, pottery, and awealth of stone tools (Fig. 9B), and record a transitionfrom hunter-gatherers to a settled mode of life, possiblywith incipient agriculture. Stone slabs were usedfor hut floors, but most of the silica implements camefrom the Kaimur Hills, suggesting periodic migrationbetween upland and lowland sites. Although radiocarbondates are lacking for the site, Sharma et al. (1980) usedlocal and regional evidence to suggest occupation over a10,000 yr period from the 17th to 7th millennium BC�18–19 kyr to 8–9 kyr BP). Mesolithic sites in central andwestern India are generally dated between 10 and 2 kyr BP(Misra, 2001). Sections recorded by Williams and Clarke(1995) close to the excavation site yielded MiddlePaleolithic to Mesolithic artifacts. The Mesolithic periodrepresents the first human colonization of the Gangaplains, where many sites are recorded (Sharma et al., 1980;Misra, 2001).At Mahagara, a Neolithic settlement appears to have
occupied a hollow formed by erosion of the topmostcemented gravel bed (Fig. 3A), close to the river bluff.A 2.6m deep excavation unearthed huts, pottery, tools andanimal bones, with evidence for food storage and cookingvessels. Hoof prints indicated that animals were domes-ticated and occupied pens. Layers of sand, reworkedkankar, and shells suggest that the settlement wasperiodically flooded, either from the river or from overlandflow. If the site rests against the gravel bed, a period ofgully erosion probably preceded settlement. The reworkedcarbonate gravel yielded late Upper Paleolithic artifactsand non-geometric microliths, similar to those in Layer 10at Chopani-Mando (Williams and Clarke, 1995), althoughno Paleolithic settlement was identified during excavation.Two shell samples from the gravel yielded dates of11.2 kyr BP, providing an upper age limit for the Neolithicsite, with somewhat older charcoal also present (Table 2).A Neolithic midden exposed by gully erosion at a lowerelevation yielded charcoal with dates of 3.2–4 kyr BP(Index Trench in Table 2). These may date the Neolithicactivity, although Sharma et al. (1980) noted some
ARTICLE IN PRESS
Mahagara(N)
Koldihwa(N-C)
Chillahia
Inset terrace
Rockoutcrops
Aba
ndon
ed
cour
se
Scarp(Water fall)
3m deeprock-flooredchannel+bank outcrop
Lakes/swamp
Lowterrace
2 m of strata(incised)on bedrock
Belan R.
(Pre
sent c
ourse)
BelanMainSection
Chopani-Mando(UP-M)
C- Chalcolithic
N = Neolithic
M + Mesolithic
UP = UpperPaleolithic
Seoti (
W)
Seoti (E)
0
km
2
Fig. 9. (A) Map of Belan river and archeological sites. The abandonment
of the old Belan course may represent relocation of the channel into the
Seoti East tributary (Srivastava, 1982) which may previously have joined
the Belan near Mahagara. (B) Mesolithic tools at surface at Chopani-
Mando, resting on Vindhyan quartzite slab.
M.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 407
confusion about the stratigraphic setting of the midden.According to Misra (2001), Neolithic sites in India aregenerally dated from about the middle of the thirdmillennium BC to the beginning of the first millenniumBC—from �4.5 to 3 kyr BP.
At Koldihwa across the Belan from Mahagara, Neo-lithic, Chalcolithic, and Iron Age sites were discovered.Three charcoal dates from the Neolithic site are in the6.95–10.25 kyr BP range, and were considered reliable bySharma et al. (1980), although they would seem to be old incomparison with sites elsewhere in India. Neolithic potteryfrom the site yields evidence that the inhabitants used adomesticated variety of rice.
The occupation at Chopani-Mando appears to span theperiod from the LGM through a phase of climaticinstability and monsoon intensification. This would havebeen a difficult time for stone-age peoples, and acute
scarcity of food and water in the Vindhyan uplands in thesummers may explain their northward migration to alowland site (Sharma et al., 1980). By the establishment ofthe younger Mahagara and Koldihwa settlements, themonsoon may have been near its peak, with enhancedflooding suitable for the development of agriculture,including rice cultivation. These conclusions are in accordwith a growing body of evidence that climatic eventsgreatly affected patterns of early human settlement acrossEurope and Asia (Staubwasser and Weiss, 2006; Weningeret al., 2006; Yancheva et al., 2007).
7.2. Channel abandonment and settlement shift
Two aspects of the sites deserve comment. One is theabandonment of the former Belan channel at Chopani-Mando. The second is the shift in settlement fromChopani-Mando to Mahagara and Koldihwa between theMesolithic and Neolithic periods. We suggest that the twoaspects are related.The abandoned channel intersects the modern Belan
close to the Belan Main Section (Fig. 1C). Anecdotalinformation indicates that there is some flow in the channelduring the monsoon period. The abandoned channelappears to be a continuation of a large meander loop onthe present river (Fig. 9A), supporting the inference thatthe abandoned channel was a former Belan course; it isreferred to below as the ‘‘old Belan’’. Downstream fromthe intersection, the old Belan enters a rocky ‘‘gorge’’about 3m deep with projecting quartzite ledges (Fig. 5E),with the Chopani-Mando site on the southern bank. Belowa small scarp (which would form a waterfall during flow),the old channel enters a broader reach until it rejoins thepresent Belan at Mahagara. The Chopani-Mando sitewould have been particularly suitable for stone-agepeoples, with flowing water, a waterfall, a flat terrace areafor settlement, and the availability of stone from the gorgefor hut floors and anvils.Srivastava (1982) suggested that the old Belan was
abandoned as the river shifted into the course of itstributary the Seoti (E branch), which would previouslyhave joined the Belan near Mahagara. This is a reasonablesuggestion, although the presence of channel fills in theChillahia section implies that large alluvial channels hadpreviously occupied this course. As a consequence of valleyaggradation during MIS 5–3, the Belan channel floorwould have been well above bedrock along much of itsreach, and a shift into the Seoti would have involvedincising through alluvium near the Belan Main Section,rather than into bedrock. The old Belan channel—lockedin place on a high bedrock area at Chopani-Mando—would have been unable to deepen its course to maintain agradient advantage.Two possible timings are suggested for the change of
course. Firstly, topmost alluvium near the Belan MainSection yielded a date of 1673 kyr BP (Williams et al.,2006), which should pre-date incision. Incision and channel
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410408
relocation could have taken place during early monsoonintensification, after �15–15.5 kyr BP based on the SanaiTal record. However, the Chopani-Mando site would havebeen occupied after the nearby river had been abandoned.A second possibility is that the channel shifted during alater period of monsoonal intensification, perhaps as themonsoon peaked towards 6.7 kyr BP (Sanai Tal data).A shift at this time might have lead to abandonment of theChopani-Mando settlement, with later re-establishment ofNeolithic settlements along the active river at Mahagaraand Koldihwa. This later timing is in broad accord with theage of the Neolithic sites, and perhaps with the incision ofthe terrace at Mahagara. The large inset terrace (Fig. 5F)just upstream from Mahagara, which is undated butevidently young, may reflect increased sediment supplyalong the new Belan course.
It is possible, however, that abandonment of the oldBelan was not an abrupt event but involved graduallydeclining flow over a long period. Additionally, manyother explanations for the shift in settlement locationcould be envisaged, including many social and culturalpossibilities that are not readily amenable to geologicaltesting.
8. Conclusions
The Belan River occupies a gentle bedrock valleyemerging from the Vindhyan Kaimur Hills at the southernmargin of the Ganga Plains of India. It contains aremarkable suite of archeological sites that have yielded awealth of information. Based on incised alluvial terracesnear the sites and 13 new age dates, we present asedimentological and paleoclimatic analysis for the area,along with a compilation of previous age dates. Where wehave OSL and shell radiocarbon dates from the same bed,the shell dates are commonly several thousand years older,probably reflecting reworking of shells and/or sources withradiogenically depleted carbon. Individual terraces yieldvaried ages, and terraces may have aggraded at differenttimes along the valley.
Above undated pedogenic and channel calcretes restingon bedrock, nearly 20m of alluvium was deposited frommixed-load meandering rivers. Middle Paleolithic artifactshave been recovered from these strata, for which we haveobtained dates between 85711 and 7278 kyr BP. Thesedates imply sustained fluvial activity during MarineIsotope Stage (MIS) 5, probably continuing through toMIS 3. We have no dates that record events during the LastGlacial Maximum (LGM). However, thin reworkedgravels with Upper Paleolithic artifacts, previously datedat �21–31 kyr BP, may represent declining alluviation andenhanced gully erosion of the floodplains under conditionsof reduced monsoonal activity. Alluvial successions else-where in the southern Ganga Plains testify to reduceddischarge during this period.
We record a suite of age dates for strata younger thanthe LGM. The dates were obtained from shell-rich eolian
sand within fluvial channels, as well as from mounds ofshelly sand inland from the present Belan course. Thesands span the 14–7 kyrBP period (including age uncer-tainties), corresponding to a period of climatic instabilitythat includes the Younger Dryas as the monsoon began tointensify. Although suggesting a more arid period, inaccord with evidence from elsewhere in the southern plains,the Belan eolian material could in part reflect local windaction near the Kaimur Hills. Overlying floodplain mudsreflect renewed alluviation, after which the river incisedduring peak monsoon flow.Although the ages of the archeological settlements are
not fully established from independent evidence, thefamous Mesolithic settlement of Chopani-Mando probablyspans a period of reduced monsoonal activity and climaticinstability following the LGM. Subsequent Neolithicsettlements were probably established under strongermonsoon conditions suitable for the development ofagriculture. Mesolithic habitation may have ended whena nearby bedrock channel was abandoned as the reinvigo-rated Belan cut a new course, possibly entering a formertributary. Neolithic settlements that have yielded the firstknown evidence of rice cultivation were then establisheddownstream along the new course.The study highlights the difficulty of constraining the
history of archeological sites with reliable age dates.Luminescence dates work well for long periods such asthe Middle Paleolithic, where age uncertainties are not ofgreat concern. However, the greatly increased rate ofchange in human culture from the Upper Paleolithiconwards creates difficulties in setting up a high-resolutionrecord. Problems include the indirect nature of most dateswhich constrain but do not ‘‘date’’ the human events, thelarge luminescence error ranges, and discrepancies betweendifferent methods, some of which may reflect reworking ofshells and charcoal. Nevertheless, dating of key events tiedto a precise stratigraphic framework, as well as cross-checking different dating methods, provides a reasonableframework for linking the cultural evolution of theseimportant stone-age sites to changes in climate andenvironment.
Acknowledgments
We thank Lewis Owen and an anonymous reviewer fortheir thoughtful and helpful comments on the originalmanuscript, and Jim Rose for editorial assistance. ParthaBhattacharjee and C.R. Sumesh provided field assistance,and we thank villagers in the Belan Valley for their kindassistance. N.G.R. and M.J. thank Andrew Murray atRisø National Laboratory, Denmark for his assistancewith luminescence dating, and Alex Cherkinsky providedthe AMS dates at Geochron Laboratories. M.R.G. isgrateful for financial support from a Discovery Grantprovided by the Natural Sciences and Engineering Re-search Council of Canada.
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410 409
References
Agarwal, K.K., Singh, I.B., Sharma, M., Sharma, S., 2002. Extensional
tectonic activity in the cratonward parts (peripheral bulge) of the
Ganga Plain foreland basin, India. International Journal of Earth
Science 91, 897–905.
Aitken, M.J., 1985. Thermoluminescence Dating. Academic Press,
London, UK, 359pp.
Allen, J.R.L., 1965. The sedimentation and palaeogeography of the Old
Red Sandstone of Anglesey, North Wales. Proceedings of the
Yorkshire Geological Society 35, 139–185.
Clemens, S.C., Prell, W.L., 2003. A 350,000 year summer-monsoon multi-
proxy stack from the Owen Ridge, Northern Arabian Sea. Marine
Geology 201, 35–51.
Dhir, R.P., Tandon, S.K., Sareen, B.K., Ramesh, R., Rao, T.K.G.,
Kailath, A.J., Sharma, N., 2004. Calcretes in the Thar Desert: genesis,
chronology and paleoenvironment. Proceedings Indian Academy of
Science (Earth and Planetary Science) 113, 473–515.
Folk, R.L., 1974. Petrology of Sedimentary Rocks. Austin, TX, Hemphill,
182pp.
Friend, P.F., Sinha, R., 1993. Braiding and meandering parameters. In:
Best, J.L., Bristow, C.S. (Eds.), Braided Rivers. Geological Society,
London, Special Publication, vol. 75, pp. 105–111.
Gibling, M.R., Tandon, S.K., Sinha, R., Jain, M., 2005. Discontinuity-
bounded alluvial sequences of the southern Gangetic Plains, India:
aggradation and degradation in response to monsoonal strength.
Journal of Sedimentary Research 75, 369–385.
Goodbred Jr., S.L., 2003. Response of the Ganges dispersal system to
climate change: a source-to-sink view since the last interstade.
Sedimentary Geology 162, 83–104.
Jain, M., Murray, A.S., Bøtter-Jensen, L., Wintle, A.G., 2005a. A single-
aliquot regenerative-dose method based on IR bleaching of the fast
OSL component in quartz. Radiation Measurements 39, 309–318.
Jain, M., Bøtter-Jensen, L., Murray, A.S., Denby, P.M., Tsukamoto, S.,
Gibling, M.R., 2005b. Revisiting TL: dose measurement beyond the
OSL range using SAR. Ancient TL 23, 9–24.
Ji, S., Xingqi, L., Sumin, W., Matsumoto, R., 2005. Palaeoclimatic
changes in the Qinghai Lake area during the last 18,000 years.
Quaternary International 136, 131–140.
Jin, Z.-D., Wu, Y., Zhang, X., Wang, S., 2005. Role of lateglacial to mid-
Holocene climate in catchment weathering in the central Tibetan
Plateau. Quaternary Research 63, 161–170.
Kale, V.S., Gupta, A., 2001. Introduction to Geomorphology. Orient
Longman, Calcutta, 274pp.
Machette, M.N., 1985. Calcic soils of the southwestern United States.
Geological Society of America Special Paper 203, 1–21.
Martin, R.E., Wehmiller, J.F., Harris, M.S., Liddell, W.D., 1996. Comparative
taphonomy of bivalves and foraminifera from Holocene tidal flat
sediments, Bahia la Choya, Sonora, Mexico (northern Gulf of California):
Taphonomic grades and temporal resolution. Paleobiology 22, 80–90.
Mishra, S., 1995. Chronology of the Indian Stone Age: the impact of recent
absolute and relative dating attempts. Man and Environment 20, 11–16.
Misra, V.N., 2001. Prehistoric human colonization of India. Journal of
Bioscience, Indian Academy of Sciences 26, 491–531.
Nakagawa, T., Kitagawa, H., Yasuda, Y., Tarasov, P.E., Gotanda, K.,
Sawai, Y., 2005. Pollen/event stratigraphy of the varved sediment of
Lake Suigetsu, central Japan from 15,701 to 10,217 SG vyr BP (Suigetsu
varve years before present): description, interpretation, and correlation
with other regions. Quaternary Science Reviews 24, 1691–1701.
Nash, D.J., Smith, R.F., 2003. Properties and development of channel
calcretes in a mountain catchment, Tabernas Basin, southeast Spain.
Geomorphology 50, 227–250.
Overpeck, J., Anderson, D., Trumbore, S., Prell, W.L., 1996. The
southwest Indian monsoon over the last 18,000 years. Climate
Dynamics 12, 213–225.
Owen, L.A., Finkel, R.C., Barnard, P.L., Haizhou, M., Asahi, K., Caffee,
M.W., Derbyshire, E., 2005. Climatic and topographic controls on the
style and timing of Late Quaternary glaciation throughout Tibet and
the Himalaya defined by 10Be cosmogenic radionuclide surface
exposure dating. Quaternary Science Reviews 24, 1391–1411.
Owen, L.A., Bright, J., Finkel, R.C., Jaiswal, M.K., Kaufman, D.S.,
Mahan, S., Radtke, U., Schneider, J.S., Sharp, W., Singhvi, A.K.,
Warren, C.N., 2007. Numerical dating of a Late Quaternary spit-
shoreline complex at the northern end of Silver Lake playa, Mojave
Desert, California: a comparison of the applicability of radiocarbon,
luminescence, terrestrial cosmogenic nuclide, electron spin resonance,
U-series and amino acid racemization methods. Quaternary Interna-
tional 166, 87–110.
Pal, J.N., Williams, M.A.J., Jaiswal, M., Singhvi, A.K., 2004. Infra red
stimulated luminescence ages for prehistoric cultures in the Son and
Belan Valleys, north central India. Journal of Interdisciplinary Studies
in History and Archaeology 1, 51–62.
Phadtare, N.R., 2000. Sharp decrease in summer monsoon strength
4000–3500 cal yrB.P. in the central Higher Himalaya of India based on
pollen evidence from alpine peat. Quaternary Research 53, 122–129.
Porter, S.C., 2001. Chinese loess record of monsoon climate during the last
glacial–interglacial cycle. Earth-Science Reviews 54, 115–128.
Prasad, S., Enzel, Y., 2006. Holocene paleoclimates of India. Quaternary
Research 66, 442–453.
Prell, W.L., Kutzbach, J.E., 1987. Monsoon variability over the past
150,000 years. Journal of Geophysical Research 92, 8411–8425.
Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Blackwell,
P.G., Buck, C.E., Burr, G.S., Cutler, K.B., Damon, P.E., Edwards,
R.L., Fairbanks, R.G., Friedrich, M., Guilderson, T.P., Herring, C.,
Hughen, K.A., Kromer, B., McCormac, F.G., Manning, S.W.,
Ramsey, C.B., Reimer, R.W., Remmele, S., Southon, J.R., Stuiver,
M., Talamo, S., Taylor, F.W., van der Plicht, J., Weyhenmeyer, C.E.,
2004. IntCal04 Terrestrial radiocarbon age calibration, 0–26 cal k-
yrBP. Radiocarbon 46, 1029–1058.
Resig, J.M., 2004. Age and preservation of Amphistegina (foraminifera) in
Hawaiian beach sand: implication for sand turnover rate and resource
renewal. Marine Micropaleontology 50, 225–236.
Sharma, G.R., Misra, V.D., Mandal, D., Misra, B.B., Pal, J.N., 1980.
Beginnings of Agriculture. Abinash Prakashan, Allahabad, 320pp.
Sharma, S., Joachimski, M., Sharma, M., Tobschall, H.J., Singh, I.B.,
Sharma, C., Chauhan, M.S., Morgenroth, G., 2004. Lateglacial and
Holocene environmental changes in Ganga plain, Northern India.
Quaternary Science Reviews 23, 145–159.
Sinha, R., Tandon, S.K., Gibling, M.R., Bhattacharjee, P.S., Dasgupta,
A.S., 2005. Late Quaternary geology and alluvial stratigraphy of the
Ganga basin. Himalayan Geology 26, 223–240.
Sinha, R., Tandon, S.K., Sanyal, P., Gibling, M.R., Stuben, D., Berner,
Z., Ghazanfari, P., 2006. Calcretes from a Late Quaternary interfluve
in the Ganga Plains, India. Carbonate types and isotopic systems in a
monsoonal setting. Palaeogeography, Palaeoclimatology, Palaeoecol-
ogy 242, 214–239.
Srivastava, K.M., 1982. New Era of Indian Archaeology. Cosmo
Publications, New Delhi, 193pp.
Srivastava, P., Juyal, N., Singhvi, A.K., Wasson, R.J., Bateman, M.D.,
2001. Luminescence chronology of river adjustment and incision of
Quaternary sediments in the alluvial plain of the Sabarmati River,
north Gujarat, India. Geomorphology 36, 217–229.
Srivastava, P., Singh, I.B., Sharma, M., Singhvi, A.K., 2003. Lumines-
cence chronometry and Late Quaternary geomorphic history of the
Ganga Plain, India. Palaeogeography, Palaeoclimatology, Palaeoecol-
ogy 197, 15–41.
Staubwasser, M., Weiss, H., 2006. Holocene climate and cultural
evolution in late prehistoric–early historic West Asia. Quaternary
Research 66, 372–387.
Tandon, S.K., Sareen, B.K., Rao, M.S., Singhvi, A.K., 1997. Aggradation
history and luminescence chronology of late Quaternary semi-arid
sequences of the Sabarmati basin, Gujarat, Western India. Palaeogeo-
graphy, Palaeoclimatology, Palaeoecology 128, 339–357.
Tandon, S.K., Gibling, M.R., Sinha, R., Singh, V., Ghazanfari, P.,
Dasgupta, A.S., Jain, M., Jain, V., 2006. Alluvial valleys of the
ARTICLE IN PRESSM.R. Gibling et al. / Quaternary Science Reviews 27 (2008) 391–410410
Gangetic Plains, India: timing and causes of incision. In: Dalrymple,
R.D., Leckie, D.A., Tillman, R. (Eds.), Incised Valleys in Time and
Space. SEPM Special Publication, vol. 85. Tulsa, Oklahoma, USA, pp.
15–35.
Thamban, M., Purnachandra Rao, V., Schneider, R.R., Grootes, P.M.,
2001. Glacial to Holocene fluctuations in hydrography and productiv-
ity along the southwestern continental margin of India. Palaeogeo-
graphy, Palaeoclimatology, Palaeoecology 165, 113–127.
van Campo, E., Gasse, F., 1993. Pollen- and diatom-inferred climatic and
hydrological changes in Sumxi Co Basin (Western Tibet) since
13,000 yrB.P. Quaternary Research 39, 300–313.
Weninger, B., Alram-Stern, E., Bauer, E., Clare, L., Danzeglocke, U.,
Joris, O., Kubatzki, C., Rollefson, G., Todorova, H., van Andel, T.,
2006. Climate forcing due to the 8200 cal yrBP event observed at Early
Neolithic sites in the eastern Mediterranean. Quaternary Research 66,
401–420.
Williams, M.A.J., Clarke, M.F., 1984. Late Quaternary environments in
north-central India. Nature 308, 633–635.
Williams, M.A.J., Clarke, M.F., 1995. Quaternary geology and prehistoric
environments in the Son and Belan valleys, north central India.
Memoir, Geological Society of India 32, 282–308.
Williams, M.A.J., Pal, J.N., Jaiswal, M., Singhvi, A.K., 2006. River
response to Quaternary climatic fluctuations: evidence from the Son
and Belan valleys, north-central India. Quaternary Science Reviews 25,
2619–2631.
Wintle, A.G., Murray, A.S., 2006. A review of quartz optically stimulated
luminescence characteristics and their relevance in single-aliquot
regeneration dating protocols. Radiation Measurements 41, 369–391.
Yancheva, G., Nowaczyk, N.R., Mingram, J., Dulski, P., Schettler, G.,
Negendank, J.F.W., Liu, J., Sigman, D.M., Peterson, L.C., Haug,
G.H., 2007. Influence of the intertropical convergence zone on the East
Asian monsoon. Nature 445, 74–77.